Method and device of measuring infrared spectral characteristics of moving target

1. A method, comprising: performing following steps through a computer: (1) establishing a multi-dimensional and multi-scale model with infrared spectral features of an object-space target, and extracting an object-space region of interest (ROI) measurement model; (2) performing target detection on an actually measured infrared image captured by an imaging infrared spectrometer, and identifying position information for each ROI of a target; tracking the target, to obtain the target's pixel differences between two frames, and a moving direction of the target, and according to the target's pixel differences between the two frames, performing motion compensation for the target, wherein (2) comprises: (2.1) performing multi-threaded operations to an input actually measured infrared image, comprising: a 1 st thread: performing hyperpixel segmentation to obtain sky background, ground background, target regions, etc., and based on the segmentation result and measurement of area and grayscale features of a target, identifying background regions and taking as a negative sample; a 2 nd thread: extracting full-image Histogram of Oriented Gradient (HOG) features, and distinguish background from the target regions according to a slide-window method, thus obtaining a suspected target and taking as a positive sample; a 3 rd thread: using a full convolutional neural network to detect an input image to obtain a target, and taking it as a positive sample; (2.2) inputting the results obtained by the above threads into a pre-trained support-vector-machine (SVM) classifier, to obtain position information (x, y) of the image of the target; (2.3) creating a Gaussian pyramid by using the image of the target, to obtain multi-scale information of the image and thereafter input it into the trained convolutional neural network (CNN) to obtain pixel differences (Vx, Vy) of respective ROIs of the target with respect to the center position of the target; and (2.4) detecting and obtaining position information of two-frame images of the target according to (2.2) and processing, to obtain the target's frame differences as well as the moving target's direction information from the two-frame images; based on the target's pixel differences between the two frames obtained in (2.3), performing motion compensation for the target; (3) scanning the target identified in (2), and after successfully capturing an image of the target being tracked, controlling an inner framework to point to each target of interest, and according to moving-direction information of the target, performing N-pixel-size motion in a direction shifted by 90° with respect to the moving direction, activating a spectrum measuring module, recording distance information between the measuring device and the target, measuring azimuth information, scale information and time-dimension information, and inputting the information into the multi-dimensional and multi-scale model obtained in (1).

2. The method of claim 1 , wherein (1) further comprises: (1.1) establishing a three-dimensional model for a target to be measured; (1.2) from the three-dimensional model, determining a ROI section of the target to be measured, and performing material classification to the three-dimensional model of the above-mentioned section to determine a radiation source; (1.3) measuring infrared spectral characteristics of the radiation source, to obtain object-space infrared-spectral-characteristic distribution at a specific angle.

4. A measurement device for implementing the method of claim 1 , the device comprising: an industrial computer, a rotary mirror, a beam splitter, a medium-wave lens, a long-wave lens, a non-imaging infrared spectrum measuring unit, and a long-wave infrared imaging unit; wherein: a control interface of the industrial computer is connected to the rotary mirror; the medium-wave lens is mounted on the non-imaging infrared spectrum measuring unit; an output end of the non-imaging infrared spectrum measuring unit is connected to an input end of the industrial computer; the long-wave lens is mounted on the long-wave infrared imaging unit; and an output end of the long-wave infrared imaging unit is connected to the input end of the industrial computer.

5. The device of claim 4 , wherein the rotary mirror adopts a four-framework servo control and comprises: a reflective mirror, an inner pitch framework, an inner azimuth framework, an outer pitch framework, and an outer azimuth framework, which are sequentially arranged from inside to outside.